![[Lk].Zano](https://steamcdn-a.akamaihd.net/steamcommunity/public/images/avatars/b9/b98e01ce6ef4a5e926a44cdd1564d0e8e5fbfa36_full.jpg "[Lk].Zano")
I would say around 2-2.5k. But if you're talking about average mmr on this forum, it's 6k at least.
The mean MMR is 2250... that's what most people think of for the word "average". Half of all dota players are above 2250 and half are below it.
If you want to look at only dota players serious enough to make a team and play games as a team... The mean tMMR is 3020. It has been slightly over 3k all year long. But only 1.4% of dota players actually make teams and use team matching. This biases the scores so that they include most of the best players and almost none of the bad and mediocre players.
Nevertheless there are still some people who play team matches all the way down to below 1k MMR. Teamwork and skill are only very loosely connected. You will find very cooperative people who are quite bad at dota and extremely high skill players who are so selfish no one wants to play with them.
3.3k is technically a top 10% score.... although it is so little respected on dotabuff people are outraged and incredulous whenever I mention that fact. While 3.3k is better than 90% of dota players this is because ordinary average players are casuals who can't lasthit, don't look at the map, don't try hard to win, don't play often enough to practice effectively, and don't have much game knowledge.
There is very large skill gap from 2.2k to 2.6k and from 2.6k to 3k, and from 3k to 3.4k etc... In fact 400 points in MMR is so large that if you beat a team with an average 400 points below yours you will get no points for the win. Usually Valve tries hard to make the teams average MMR equal but occasionally they are not. For ranked games Valve has switched from TSR to ELO type calculations so if you don't get +/- 25 points for a game it means the average MMR for the teams was not equal.
Despite this, people have a huge range of skill on different heroes. If you were to play only your very best hero if would likely end up at 50% win leveling out at an MMR about 1000 points higher than if you were to play only your worst hero. Once you have played a lot of games it becomes easy to see that some of you heroes are dramatically more likely to win than others. That is not the heroes intrinsic power... that is your relative skill on the various heroes.
Compare your winrate and other stats for a hero to the average score for pub games and you will see large differences for your best and worst heroes. This does not correspond to most played. People often play a lot of games on a hero they are not playing correctly, or skillfully and so lose plenty of games. More practice does not always fix it.
Pretty much everyone at 2500+ knows what each skill of each hero do. I'm sure there are plenty of people below that.
There are too many standards for what's "average".
Dotabuff standard says 4k-4.5k.
Valve standards say around 2.3k.
The "most of my friends have ___ mmr" standard for me is 3.5k.
Personally, I think what Relentless(valve's stats) was saying should be the standard for what's average. Because it accounts for all kinds of players, from clueless noop to professional tier, regardless of age & region.
Valve standards say "There's retarded, there's bad, there's not so bad, there's average, there's good, there's very good, then there's pro tier" while in dotabuff standards, it's you're either bad or "pro".
@way2high
Depends who you compare with. If you compare with Mushi, then you'd pe scrub even with 5k MMR. Check this thread: http://dotabuff.com/topics/2014-01-24-list-of-people-at-high-mmr-that-are-garbage
Relentless
4 hours ago
The mean MMR is 2250... that's what most people think of for the word "average". Half of all dota players are above 2250 and half are below it.
You confuse mean and median.
Do you go up higher in MMR if you have played less games? If not then wouldn't the first 10 games or so be extremely important?
I guess I did confuse mean and median a bit... in this case they are the same thing (or very close) by design.
Yes the first ten games you can gain as much as 2k MMR. You start with 2k MMR, but if you get very high KDA and win the first ten games you can go up to about 4k immediately. Then all games are worth about 25 points.
^
not sure how true that is because i won 3 lost 7 and got placed at 3.9k
and my friend won 8 lost 2 and was placed at 3.4k
both our kdas were around 2
@Relentless
Damn. If that's true, I feel bad about rushing/throwing my calibration games. I almost lost all (if not all, definitely around 80% lost) my calibration games on both party and solo and I got 4503 for both.
I'm not sure about the calibration games. I mean that you start a new account with 2k. Calibration moves you up and down a lot also, but not as much as the anti-smurf... maybe 500 MMR max for calibration moves. But I've seen so few of those I'm just guessing.
Ah yes. I remember someone saying there's hidden MMR for old accounts so yeah. Definitely different than new ones.
Concede the MMR distribution for ranked will be nearly identical to non-ranked MMR just because the numbers of players are very large. It can't move much. Some individuals will go up and down a lot. I have gained about 700 MMR since calibration did not go well and started me low. But the whole system won't move.
If 4k to 4.5k is avg then someone like purge who is 4.79k(or whatever it is) is good enough to join a korean team? So the korean scene is not as strong?
@concede i got this info from some official like a week ago, dont remember whether it was a forum or the volvo blog thing, but its between 2250 and 2500
Relentless - while you make some points, you are missing out on something very important. Simply because the population is large does not mean anything about the mean of a population changing when a significant portion is removed, and a majority of the removed portion is from one arm of the distribution.
The MMR distribution will change because the large number and characteristics of players will change by a LOT. I think a large amount of the lower ranked players will never play ranked MM, or not play all of their calibration matches. This could be between 25-50% of total players, and maybe say 75% of the players who dont play rmm are from the lowest end of the unranked MMR curve.
This results in the mean (average) mmr falling to the right (above) of the median mmr. Before, the median and mean would be approximately equal. The unranked mmr distribution is approximately normal (assumption, but most populations are at least semi-normal), and the ranked distribution would be skewed right - this means the the mean of the latter distribution will be above the former, depending on the portion of players removed from the pool and the percentage of those that were below the mean of the original population.
There is also a self selection tendency to take into account, which also contributes to an upwards skew. In general, self-selection results in a population skewing towards the characteristic which is self selected - ie. people who sign up for a dating site have a higher chance of finding a relationship/getting married. People who play ranked matchmaking are likely to be better players.
TLDR: the average mmr will increase, depending on the size and proportion of lower rated players who dont play rmm. The numbers are large, as you said, but a very large amount of players will not play rmm (most of them being of lower mmrs) and thus the average will increase, significantly, ad the shape of the distribution is skewed upwards. The ranked distribution will be much different than unranked
Remember the American Pie rule of 3 (guy version)? The same rule applies when you ask someone there MMR
korean scene is similar to special olympics
wait a couple of years until koreans know how to play dota andy ou'll see players like purge, eosin etc deepthroat their D
Do you think that Korea - starting 5-7 years behind other pro teams, can catch really catch up while pro teams who are on a different level now also continue to improve? I am not so sure myself
57% winrate, 3.5-4 KDA overall average and Volvo gives me 3600 solo MMR. And I am still getting people walk into enemy jungle solo when nobody on the minimap, getting supports with no wards in my team on half of the games. I don't know the average MMR but you have to get at least 4.5k to get decent teammates in your team imo.
you have approximately 25 net wins. 25 wins x 25 mmr per game +3000 is approximately where you are at. Holy shit math is so hard!
@Reguardz
you buffed your stats making a smurf account. You do not deserve more than 3k MM
back on topic MM is just a rough estimate on where you are suitable to play without going 0-20 every game.
2250 is from December 10, 2013 http://de.dota2.com/2013/12/matchmaking-2/
maybe it's different now ?
sigh. that was unranked. in ranked, many of the below average players are not in the distribution (dont play ranked). Thus the average will be higher
Please sign in to post comments.
What's considered average MMR?